Overview
The Algorithms module provides machine learning capabilities for scoring, experimentation, and model management.Scoring Models
Train and deploy models that score offers for individual customers. Three scoring engines are available:| Engine | Description | Use Case |
|---|---|---|
| Scorecard | Weighted attribute scoring with configurable bins | Simple, interpretable scoring |
| Bayesian | Bayesian probability model | When you have prior beliefs about conversion rates |
| Gradient Boosted | ML-based scoring with feature importance | High-accuracy scoring with sufficient training data |
Experiments
Run controlled experiments to measure the impact of your decisioning changes:- Holdout groups — Deterministic assignment ensures customers stay in control/treatment
- Uplift calculation — Two-proportion z-test with confidence intervals
- Metrics tracking — Conversion rate, revenue per decision, engagement rate
Model Management
- Version tracking for all models
- A/B testing between model versions
- Performance monitoring dashboards
- Automatic retraining triggers